Vision-Based Robotic Solution for Wire Insertion With an Assigned Label Orientation

This paper tackles the problem of wire insertion in switchgear assembly according to current regulations. In particular, connections require that label has be oriented facing up order simplify and speed testing maintenance switchgear. The proposed approach exploits a priori knowledge scenario with calibrated RGB camera robotic arm estimate both end pose position. procedure combines several techniques (gradient base, trained classifier stereo vision) elaborate standard images extract some features related its shape label. Specific frames are fixed estimated then used correctly complete task by using system. Experiments reported verify effectiveness approach.